The Neurotoxicity Hypothesis of Cognition, Risk-Taking and Violence

Kevin Drum at Mother Jones has been on a tear for years about what he calls the ‘lead-crime hypothesis.’ This is the intriguing idea that the crime wave of the late-20th century in the United States was caused by industrial and automobile childhood lead exposures at midcentury. The basic pattern that the hypothesis was meant to explain was the rise and fall of the rate of violent crimes in the United States.


It is an intriguing hypothesis. Is it right? I was astonished to discover that, not only is it right, the specific pathways through which childhood lead exposures affect large-scale behavioral patterns in society are well-understood among health scientists. The basic picture is that exposure to lead during the growth period from conception to adulthood arrests the development of the prefrontal cortex — the “CEO” of the brain. Specifically, the more exposure a child has to lead in paint, soil and air, the greater the insult to her cognitive function. In turn, impaired functioning of the prefrontal cortex strongly predicts poor impulse control, antisocial behavior, mental disability, poor academic performance, and all that follows from these mediators.

Some of the most striking evidence was published in 1998 when Raine and other brain scientists published their study of cold-blooded and hot-blooded murderers. Specifically, they compared the pattern of brain activity in impulsive murderers and those who had committed preplanned murders with each other and a control group. Already by 1988, it had been established that the subcortical region of the brain’s right hemisphere is involved in the generation of aggressive feelings and behavior, while the prefrontal cortex is responsible for disciplining these violent impulses. Raine and his coauthors showed that glucose metabolism was impaired in the prefrontal cortex of both impulsive affective murderers and calculating predatory murderers, and metabolism was more pronounced in the subcortex of both — but the difference was more pronounced in the impulsive affective group.





Because the theory is so tight — ya can’t fiddle with it at all — it generates a whole series of predictions which can be tested against reality. In a series of papers, results later collected in his Lucifer Curves: The Legacy of Lead Poisoning, Rick Nevin documented the overwhelming scientific evidence for the neurotoxicity hypothesis.

US violent crime
Source: Nevin (2016).

More recently, Jessica W. Reyes investigated the effect of early childhood lead exposure on behavioral problems from childhood to early adulthood using the National Longitudinal Surveys of Youth. She documented a strong association between teen pregnancy rates and lead exposure in the panel data; and a significant but weaker association of lead exposure with violence and the behavioral problems index.

Source: Reyes (2014).

In 2012, Mielke and Zahran showed that the concentration of lead in the air strongly predicts 22-year ahead aggravated assault rate in all the US cities for which they could find data. The pattern for New Orleans:

Mielke and Zahran 2012a

… is typical of US cities:

Mielke and Zahran 2012b

Nevin has shown that the pattern holds internationally as well. Different countries phased out lead gasoline and lead paint at different times. In all of them, you see the same lagged correlation structure. The evidence could not be more compelling if it tried.

But you know me, I like to check things by hand. We obtain indicators of risky behavior at the county level from Social Explorer and blood lead levels reported by the CDC. We find that blood lead levels are very strongly associated with premature age-adjusted mortality in the cross-section of US counties (r=0.40, P<0.001). Note that our feature is very noisy since the real signal is in cohort-specific lead exposures over the past few decades, while we are using exposure data from 2012. It is extraordinary that the association is so strong anyway.


The association is somewhat strong for teen birth rate:


It is very strong for smoking:Smoking

The effect of childhood lead exposure on life-long risk-taking is so strong that, our weak proxy for lead exposure is strongly associated with a very important overall measure of health outcomes, years of potential life lost (YPLL):


And it is not just risk-taking. Childhood lead exposure has a very strong effect on cognitive ability. In their 2018 review of the literature, “Lead (Pb) neurotoxicology and cognition,” Maria, Hill and Kline write:

Many studies have demonstrated an adverse impact of Pb on cognitive functioning. Pb has emerged as an independent risk factor for intelligence quotient (IQ) compromise. Pb-related IQ compromise has been demonstrated when a wide variety of potentially confounding variables have been controlled for including: sex, birth weight, iron (Fe) status, maternal IQ, years of education, race, maternal tobacco use during pregnancy, parents’ smoking status, whether or not there was exposure to cocaine or other drugs in utero, prenatal alcohol use, socioeconomic status, yearly household income, maternal age at delivery, quality of the home environment, number of prenatal medical visits, birth order, parity, infant head circumference, breastfed vs. bottle fed, duration of breast feeding, maternal age, mother’s marital status at the time of child’s birth, and whether biological parents live together or apart.

They note that “the heavy metal most studied as a neurotoxic agent is almost certainly Pb [lead]. Many studies have investigated the relationship between Pb exposure in early childhood and IQ (i.e., Baghurst et al., 1992; Bellinger et al., 1992; Canfield, Henderson, et al., 2003; Chiodo, Jacobson, & Jacobson, 2004; Cooney, Bell, McBride, & Carter, 1989a; Dietrich, Berger, Succop, Hammond, & Bornschein, 1993; Ernhart, Morrow-Tlucak, Wolf, Super, & Drotar, 1989; Fergusson, Fergusson, Horwood, & Kinzett, 1988; Lanphear et al., 2005; Lucchini et al., 2012; Min et al., 2009; Schnaas et al., 2000; Surkan et al., 2007; Wasserman et al., 1997; Wasserman et al., 2000; Wolf, Jimenez, & Lozoff, 1994).”

What did all these scholars discover?

Several found that “blood Pb concentrations >10 μg/dl are associated with adverse outcomes on measures of intellectual functioning and social behavior.” In the 1970s, CDC had regarded 60 μg/dl as the “level of concern.” More recently, it has abandoned the practice as evidence has surfaced that even very low levels of exposure is harmful. Low-level exposures have been shown to impact cognitive ability.

Needleman, Constantine, and Gatsonis (1990) published one of the first meta-analyses examining low-level Pb [lead] exposure in relation to its effects on children’s IQ. After statistically combining the results of 12 Pb exposure studies utilizing multiple regression, the authors found a significant negative association between low-level Pb exposure and IQ in 11 out of 12 studies utilizing either BLLs or bone Pb levels as the independent variable.

Maria, Hill and Kline report that “a decrease of 6.2-7.4 IQ points has been observed when BLL [blood lead level] is increased from 1 μg/dl to 10 μg/dl, while IQ point decline is roughly 4.6 points per every 10 μg/dl in BLL above 10 μg/dl.” In a large meta-analysis, Lanpear et al. (2005) found “an inverse relationship between BLL and IQ, such that the estimated IQ loss was 3.9 points from 2.4–10 μg/dl (Lanphear et al., 2005). Furthermore, BLLs of <7.5 μg/dl were associated with greater relative loss of IQ points versus BLLs of >7.5 μg/dl, corroborating data that demonstrated the adverse neurocognitive effects of Pb at low blood level concentrations. In other words, lower BLLs were found to be associated with steeper decrement in IQ by comparison with Pb-related IQ decrement at higher concentrations.” Surkan et al. (2007) found that “children with 5–10 μg/dl had 5.0 points lower IQ scores than children with BLLs of 1–2 μg/dl.”

The basic take-away is that each microgram per deciliter of lead concentration in the blood predicts a difference of the order of a 1 point difference in IQ.

Maria, Hill and Kline go on to report the results of dozens of studies in other countries, longitudinal studies, studies of association of lead exposure with mental disability, and so on and so forth. The bottom line is that lead exposure is a very strong predictor of cognitive ability even at very low levels of exposure. In fact, noting that mean lead levels in the US had increased from preindustrial levels of 0.02 μg/dl to 21.4 μg/dl by the late-1960s and have since fallen to 4.6 μg/dl, Kaufman et al. (2014) argue that lead exposure by itself can account for most of the Flynn effect.

Meanwhile, Cassidy-Bushrow et al. (2018) find that the concentration of lead in the teeth of African-American children is twice as high as white kids — most of it prenatal.

Cassidy-Bushrow (2018)

As Nevin explains in Lucifer Curves, “African-Americans now account for almost 40 percent of state and federal prison inmates and 42 percent of all prisoners of death row.” This “racial disparity is causal effect of housing segregation that resulted in African-Americans occupying 56% of substandard central city housing in 1960, and accounting for 48% percent of all preschool children with blood lead over 30 [μg/dl] in the late-1970s.”

Nevin BWgap

The differences are easily large enough to account for the entire black-white difference in test scores — that Rushton and Jenson (2005) estimated at 8 IQ points and Murray and Hernstein (1994) reported at 16 IQ points. The difference in the estimates of the two pairs of racialists is not a coincidence. The black-white gap has been narrowing at the same time and for the same reason that the crime rate gap between big cities and small cities is narrowing: both blacks and big cities were more exposed to lead during the midcentury passage than whites and small cities.

Crazy fucking thing isn’t it? You really have to pay attention to the details. It is, of course, stupid to fuss over raw differences in population means, as if they were anything other than explananda crying out for explanation. It is also a warning against jumping at conclusions generally. The international pattern suggested the supremacy of health insults due to malnutrition and disease burdens. The neurotoxicity hypothesis focuses our attention on a very specific sort of health insult — that due to a specific feature of the hockey stick of modernity as the controlling variable. Who would’ve thunk?

What is surprising is that so few people outside the specific sciences concerned know about the scale of the burden of lead exposures. But that’s irrelevant to the scientific discourse. As far as science is concerned, this is an open and shut case.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s